Artificial intelligence (AI) can be a strong ally for researchers, especially when processing huge amounts of information. In a world-first, a team in Australia successfully trained AI to extract data from millions of plant specimens stored in herbaria worldwide, gaining new insights into the impacts of climate change on plants.

The researchers used a new machine learning algorithm to process over 3,000 leaf samples and found that. They found that across the same species, leaf size doesn’t necessarily increase with increasing heat or wetness of the climate. This is surprising because on different species, leaf size does increase with heat and wetness.

Already, this is an intriguing finding. It also shows how AI could be used to better understand species and document the effects of climate change.

“Herbarium collections are amazing time capsules of plant specimens,” Will Cornwell, lead author and researcher at the University of New South Wales (UNSW), said in a statement. “Each year over 8000 specimens are added to the National Herbarium of New South Wales alone, so it’s not possible to go through things manually anymore.”

With the help of an algorithm

A herbarium is a collection of plant specimens preserved, labeled and stored in an organized way that facilitates access. They have existed at least since the 16th century. Typically, the plants are flattened, dried and mounted on uniformly sized paper. Each plant recorded has a label with information about it, such as the date it was collected.

A few years back, scientists started a campaign to move all the herbarium collections online. “To get the information about all of the incredible specimens to the scientists who are now scattered across the world, there was an effort to scan the specimens to produce high-resolution digital copies of them,” Cornwell explained in a statement.

The largest project was done at the Botanic Gardens of Sidney, with over one million plant specimens transformed into high-resolution digital images. Once completed, the researchers decided to take it further. They created an algorithm that could be trained to detect and measure the size of leaves of scanned samples for two plant genera.

They focused on Ficus, a genus of 850 species of woody trees, shrubs and vines, and Syzygium, generally known as lillipillies, brush cherries or satinas. The process teaches the AI to see and identify the components of a plant in the same way a human would. “We basically taught the computer to locate the leaves and then measure the size,” Cornwell said.

The algorithm was applied to analyze the relationship between leaf size and climate. A general rule in the botanical world is that in wetter climates, the leaves of plants are bigger compared to drier climates. While this pattern was visible between different plant species, the same correlation isn’t seen within a single species across the globe.

This is likely because a different process, known as gene flow, is working within species. That process weakens plant adaptation on a local scale and may be preventing the leaf size-climate relationship from developing within species. The AI model used provided sufficient accuracy to look at the links between leaf trains and climate, the researchers said.

“But because the world is changing quite fast, and there is so much data, these kinds of machine learning methods can be used to effectively document climate change effects,” Cornwell. Algorithms could also be trained to spot trends that might not be obvious to researchers, leading to new insights into how plants could respond to the effects of climate change.

The study was published in the American Journal of Botany.